1. Field of the Invention
This invention relates to an image processing method and an image processing apparatus suitably applied to an image reading apparatus such as an OCR (Optical Character Reader) or a facsimile, and more particularly to an image processing method and an image processing apparatus wherein a video signal (density value signal) from an optical sensor such as a CCD (Charge Coupled Device) is converted into a digital signal and the digital signal is processed by binary digitization processing to obtain binary values of white and black.
2. Description of the Related Art
In recent years, it is a common practice with an image reading apparatus such as a copying machine or a facsimile to use and scan an optical sensor such as a CCD in a main scanning direction (for example, a lateral direction of an original) and a sub scanning direction (for example, a longitudinal direction of an original) and process the thus read image data by digital processing.
When the apparatus of the type described is used to read a document or a drawing on which a longitudinal or lateral straight line such as, for example, a ruled line shown in FIG. 15, if the line is inclined a little relative to the main scanning direction or the sub scanning direction due to inclined arrangement of the original as seen in FIG. 16, then where the line spans over two picture elements in the main scanning direction or the sub scanning direction, the density value detected by the optical sensor is separated between the two picture elements as at a portion P1 or P2 in FIG. 18. Consequently, the density value does not reach a predetermined threshold level, which is used upon binary digitization processing, at any of the two picture elements, and the two picture elements are determined to be both white.
Consequently, the line of the image exhibits a partial drop of drops as seen in FIGS. 17 and 18, resulting in significant visual deterioration in picture quality. It is to be noted that, in FIGS. 16 and 18, each square represents a picture element as a minimum unit for which a density value is detected by an optical sensor, and the rightward direction in FIGS. 16 and 18 is the main scanning direction while the downward direction in FIGS. 16 and 18 is the sub scanning direction.
In order to prevent such a drop of a straight line as described above, an image reading apparatus employs contour emphasis processing by MTF (Modulation Transfer Function) correction. In the contour emphasis processing, if it is assumed that, for example, the density values of picture elements of a 3.times.3 matrix are represented by A to I as seen in FIG. 19 and the noticed picture element which makes an object for binary digitization processing is the center picture element having the density value E, the density value of the noticed picture element is given, using the density values of the eight picture elements around the picture element, by the following expression (1): EQU .vertline.E-(A+B+C+D+F+G+H+I)/8.vertline..times..alpha.+E (1)
where .alpha. is a predetermined constant.
Meanwhile, as a result of the progress of image processing apparatus in recent years, even it is possible to handle images close to natural pictures such as multi-value images or color images. However, in such practical industries as the OA (Office Automation) industry, binary images of white and black are handled mainly. This is because almost all objects for an image to be handled are characters. Although such a color as red or blue is used partially for the object of emphasis in materials for explanation, if many different colors are used, then it is less easy to recognize the image, and accordingly, there is some limitation in number of colors which may be used.
Although there is no particular problem when a document which involves some note written in red or blue is read by a person, if it is tried to read the document by an OCR or a facsimile which is designed to read a binary image of white and black, then a color of red becomes blurred whereas a color of green cannot be read. Even if a color scanner is employed, since normally a color sensor includes sensors for R, G, and B (red, green and blue) arranged successively in an order, the locations read by the individual color sensors are delicately different from each other. Consequently, such a character as a Chinese character is distorted even where the character is of the 12 point size, and cannot be used for an OCR or cannot be read readily by a facsimile.
Therefore, it is a popular measure to adjust the threshold level to be used for binary digitization processing based on a read density value as disclosed in Japanese Patent Laid-Open Application No. Showa 59-54376 or to employ contour emphasis processing in order to artificially adjust blurring or distortion.
With the image processing method described above, however, if such contour emphasis processing as described above is performed in accordance with the equation (1) above, also noise is emphasized, and accordingly, correction by a great amount cannot be performed. Meanwhile, although it may seem a promising idea to increase the resolution of the sensor, the apparatus becomes expensive and besides the amount of image data to be processed is increased. Accordingly, the idea does not provide a direct solution.
Further, when it is tried to perform, upon reading of an original printed in two colors or a like original, artificial adjustment by threshold level adjustment or contour emphasis processing, it sometimes occurs that, even if one of the two colors can be read with certainty, the other color cannot be read with certainty. Accordingly, when a document image in which two or more colors are used is read as a binary image of white and black, particularly where color characters are present discretely in a document, a sharp character image cannot be produced.